1. Introduction
Aerosols constitute a public health risk, as they can be considered one of the main factors contributing to poor air quality [
1]. Aerosols are mainly concentrated within the atmospheric boundary layer (ABL), with long-range transport layer sometimes located in the free troposphere. This heterogeneous distribution make their characterization difficult The ABL is defined as the lowest part of the atmosphere, influenced by the Earth’s surface by means of exchange of energy and moisture [
2]. It plays a critical role in air quality forecasts [
3] and greenhouse gas concentration budgets [
4] and it is mainly characterized by turbulent processes. The influence of solar radiation creates a daily evolution cycle of the layer [
5]. The cycle, in clear-sky situations, starts with the increase of ground surface temperature after sunrise, which intensifies the convection. It produces ascension of warm air masses and downward displacement of colder air masses, which creates a growing mixing layer (ML) [
6]. It is called this way because substances emitted into this layer disperse gradually horizontally and vertically due to the turbulence. When sufficient time is given and there are no sinks, the ML become completely mixed [
4]. The vertical turbulent mixing processes produces a strong aerosol gradient between the ABL and the free troposphere. When the ML is well developed, usually at the end of the daytime period, the ABL height (ABLH) is estimated similar to the ML height (MLH), as the turbulent mixing produces a nearly homogeneous distribution of aerosols along the complete layer. Later in the day, the gradual reduction of incoming solar irradiance during the early evening transition period causes a weakening of the turbulence and the convective processes. This creates a nocturnal boundary layer (NBL) close to the surface, that is stable and stratified. The remnant of the daytime ML is located above the NBL and it is called the residual layer (RL). When the sun rise again, a new ML begins to grow rapidly, eroding firstly the NBL and then entraining into the RL, producing an early morning transition. Since different surfaces respond differently to the solar heating, the development of the ABL is influenced by the surface albedo of the underlying surface [
7]. Therefore, the combined effects of the synoptic atmospheric conditions, such as atmospheric stability or wind shear, and surface characteristics such as cover, roughness or topography, determines the horizontal variations in ABL dynamics [
8]. There are a wide range of applications with high societal, economic, and health impacts, such as air quality [
9] the generation of renewable energy [
10] or numerical weather prediction [
11] that benefits from a better understanding of these ABL processes.
The traditional methods employed to characterize the ABLH, radiosounding in the frame of the World Meteorological Organization Radiosounding Global Network [
12]; present spatial and temporal limitations as they are usually launched only twice per day in most airports worldwide. This scarce observations prevent an accurate determination of the ABLH variations, both temporal and spatially, compromising their representativeness for urban and regional scales. It was documented that the calculation of the MLH based on potential temperature profile or Richardson number performed by numerical meteorological models produces more than 50% uncertainty in shallow boundary layers and 20% in deeper boundary layers due to scarcity in the input data [
12].
Continuous profiling of the entire ABL vertical extent is nowadays possible thanks to recent advances in ground-based remote-sensing technology and algorithm development [
13]. The high temporal and vertical resolution of remote sensing instruments permits a precise automatic detection of ABL sub-layer heights [
14]. Among novel remote sensing methods, a promising one is the ceilometer, a particular type of lidars operating with a single-wavelength and originally intended for cloud base height detection [
15]. Currently, ceilometers provide continuous high resolution aerosol backscatter profiles (every 15 s) with good spatial resolution (tenth of meters) and a large vertical range (up to several km ) in unattended continuous operation. Furthermore, ceilometers are usually operated in networks, such as EUMETNET EPROFILE [
16] and ICENET (Iberian Ceilometer Network, [
17]).
The ABLH retrieval algorithm for these instruments is based on the aerosols vertical profile. It assumes that aerosol concentrations are lower in the free troposphere than in the mixing layer, producing a strong negative gradient clearly observable in the backscatter profiles [
18]. The high temporal and spatial resolution allows the study of the aerosol concentration fluctuations produced by the constant interexchange of airmasses, those polluted with aerosols within the ML moving upward are exchanged with those clean moving downward from the free troposphere. A disagreement can result between the height of aerosol layers derived by lidar profiles and the radiosounding-derived MLH due to the inconsistency between the thermal profiles and the aerosol profile, especially during morning or evening transitions [
19]. ABLH retrieval algorithms take advantage of these characteristics to determine the height, such as gradient method [
20], wavelet covariance transform [
21] or edge detection method [
22]. More advanced methods extend the analysis to two dimensions (temporal and vertical) in order to guarantee temporal consistency [
23], or applies graph theory to track the diurnal evolution [
24]. A reliable new method called STRATfinder [
25] combines these last two methodologies applying a backward propagating layer, from the end of the day, in order to determine the type of layer by minimizing a cost function from the forward and backward runs. This layer attribution is normally assisted by commonly available surface measurements of radiation and temperature in order to decrease its uncertainty [
23].
However, more complex atmospheric situations can jeopardize the observable strong negative gradient in the backscatter profiles. For instance, desert dust intrusions, volcanic eruptions or forest fires [
26] can produce long-range transport of aerosols that affects the vertical distribution over the site due to the presence of lofted layer, which can either mix with the ABL or remain above it for long periods. This advection of aerosols increases the layer retrieval uncertainty, challenging any aerosol-based method. [
27]. The Iberian peninsula is located close to the Sahara desert and regularly receive desert dust events, occurring when the dust is injected and transported throughout the atmosphere over long distances. On a global scale, the apportion of desert dust have been estimated up to 40% of aerosol mass yearly injected into the troposphere [
28] and in particular Sahara desert emit half of the world atmospheric mineral dust [
29].
The main objective of this work is the assessment of the impact of Saharan dust intrusions on the ABLH using ceilometer signals, along four years period 2020-2023. The database of ABL heights of continuous measurements have been classified regarding the most frequent patterns of synoptic circulation. The weather conditions over the Iberian Peninsula were analyzed using cluster analysis of sea level pressure fields and six typical synoptic meteorological patterns (SMPs) were identified.
Section 2 describes the methodology employed including the instrumentation (ceilometer profiles) and algorithms (STRATfinder), the synoptic meteorological patterns and the methodology followed to identify Saharan dust intrusions.
Section 3 summarizes the main results obtained when the datasets are differentiated by season and by synoptic meteorological patterns and
Section 4 discuss the main findings of the work.
Author Contributions
Conceptualization, F.M.; methodology, F.M. and P.S.; software, F.M. and P.S.; validation, F.M., P.S. and M.P.; formal analysis, F.M. and P.S.; investigation, F.M., P.S. and M.P; resources, F.M., P.S. and M.P; data curation, F.M., P.S. and M.P writing—original draft preparation, F.M.; writing—review and editing, F.M., P.S. and M.P; project administration, F.M. and M.P.; funding acquisition, F.M. and M.P.. All authors have read and agreed to the published version of the manuscript.
Figure 1.
CIEMAT-Madrid site, located in the middle of the Iberian peninsula (left panel), and at the northwest of the Madrid city (middle panel), with the instrument (right panel). The satellite image at the left panel also shows a saharan dust intrusion.
Figure 1.
CIEMAT-Madrid site, located in the middle of the Iberian peninsula (left panel), and at the northwest of the Madrid city (middle panel), with the instrument (right panel). The satellite image at the left panel also shows a saharan dust intrusion.
Figure 2.
Synoptic meteorological patterns (SMP) obtained by cluster analysis of reanalysis global fields of sea level pressure at 12 UTC for the period 2001–2019. Colored areas represent atmospheric pressure measured in hPa. Cool colors are used to represent low pressures, while warm colors symbolize higher pressures. The X-axis represents longitude while the Y-axis represents latitude, both measured in degrees. North Atlantic, Europe and North Africa are depicted on the maps.
Figure 2.
Synoptic meteorological patterns (SMP) obtained by cluster analysis of reanalysis global fields of sea level pressure at 12 UTC for the period 2001–2019. Colored areas represent atmospheric pressure measured in hPa. Cool colors are used to represent low pressures, while warm colors symbolize higher pressures. The X-axis represents longitude while the Y-axis represents latitude, both measured in degrees. North Atlantic, Europe and North Africa are depicted on the maps.
Figure 3.
Quicklook of the range corrected signal (raw signal multiplied by the square of range) calibrated at 1064 nm, as color scale, and the prediction provided by the STRATfinder algorithm of the MLH (black circles) and ABLH (red crosses) at (a): 21/06/2022 and (b): 15/06/2021. The x-axis represents the time, 24 hours and the vertical axis is the height, with the color scale representing the range corrected signal.
Figure 3.
Quicklook of the range corrected signal (raw signal multiplied by the square of range) calibrated at 1064 nm, as color scale, and the prediction provided by the STRATfinder algorithm of the MLH (black circles) and ABLH (red crosses) at (a): 21/06/2022 and (b): 15/06/2021. The x-axis represents the time, 24 hours and the vertical axis is the height, with the color scale representing the range corrected signal.
Figure 4.
MLH estimations (black crosses, left y-axis) and Saharan dust load (orange circles, right y-axis) for days between January 2020 and December 2023.
Figure 4.
MLH estimations (black crosses, left y-axis) and Saharan dust load (orange circles, right y-axis) for days between January 2020 and December 2023.
Figure 5.
Boxplots of MLH for all the days separated by (a): season and (b) synoptic meteorological pattern. As usual, the red line in the middle of the boxplot represents the media of the values for that group, the box comprises the interquartile range and the top and bottom lines are the maximum and minimum values respectively. The mean have also been represented as blue circles.
Figure 5.
Boxplots of MLH for all the days separated by (a): season and (b) synoptic meteorological pattern. As usual, the red line in the middle of the boxplot represents the media of the values for that group, the box comprises the interquartile range and the top and bottom lines are the maximum and minimum values respectively. The mean have also been represented as blue circles.
Figure 6.
Seasonal distribution of cases, number of cases for each season, divided by the total number of cases assigned to that SMP, for the six synoptic meteorological patterns during the period 2020 - 2023.
Figure 6.
Seasonal distribution of cases, number of cases for each season, divided by the total number of cases assigned to that SMP, for the six synoptic meteorological patterns during the period 2020 - 2023.
Figure 7.
Boxplots of MLH for (a): Saharan and clean days and (b) High dust load, low dust load and clean days.
Figure 7.
Boxplots of MLH for (a): Saharan and clean days and (b) High dust load, low dust load and clean days.
Figure 8.
Boxplots of Saharan and clean days separated by (a): season and (b) synoptic meteorological pattern.
Figure 8.
Boxplots of Saharan and clean days separated by (a): season and (b) synoptic meteorological pattern.
Figure 9.
Distribution of clean (blue bars), low dust load (red bars) and high dust load days (orange bars) regarding the synoptic meteorological patterns for the period 2020 – 2023.
Figure 9.
Distribution of clean (blue bars), low dust load (red bars) and high dust load days (orange bars) regarding the synoptic meteorological patterns for the period 2020 – 2023.
Figure 10.
Boxplots of High dust load, Low dust load and clean days separated (a) by season and (b) by synoptic meteorological pattern.
Figure 10.
Boxplots of High dust load, Low dust load and clean days separated (a) by season and (b) by synoptic meteorological pattern.